整合個人興趣與推播公平性的廣告推薦方法

Abstract

線上廣告近年來為許多網站帶來了龐大的收益。而對網站而言，適當的廣告推薦，能夠增加收益並提高使用者對網站的滿意度。線上廣告的種類繁多，對於線上橫幅式廣告來說，通常被放置在新聞網站或是影音網站中，而其計費方式通常採用「單一廣告支付」，意味著廣告商以固定的金額，向網站租借一段固定的時間來進行廣告推播。由於網站對於每個廣告的收費是固定的，所以網站必須保證每個廣告被推播的次數也是相同的。所以，針對這樣的廣告推播條件，我們提出一個公平機制以應用在廣告推薦方法當中。 本研究提出了一個兼具推播公平性與個人興趣的廣告推薦方法。我們考量每個廣告被推播的次數，並針對使用者在網站中的三種行為進行研究，找出影響使用者喜好的主要因素、增加廣告的點擊率，進而提高廣告的商業價值。當使用者在瀏覽新聞網頁時，能夠公平且精確的推薦適當的廣告給每一個使用者。我們的實驗結果顯示，我們的方法不只能夠提高網站中所有廣告的平均點擊率，並確保每個廣告的推播次數相近且具公平性。我們的方法在考量公平原則下，展現了良好的廣告推薦成效。

Nowadays, online advertisement brings huge revenue to many websites. Appropriate ad recommendation can maximize a website’s revenue and user satisfaction. There are many kinds of types of online advertisement, we focus on the online banner ads which usually placed in a particular news website or video website. The most common paying model of banner is “pay-per-ad,” which means that the website is paid when the advertiser rent a banner during particular period. In this paying model, the website need to ensure that the ad pushed frequency of each ad on the banner is similar. Under such advertisement push rules, we present an ad-recommendation mechanism with ad push fairness. In our research, we proposed a novel method that consider both ad-push fairness and personal interest. We take every ad’s exposure times into consideration, and investigate user’s three different usage experience in the website to identify the main factors affecting the interest of user. Consequently, we can improve ad click-through rate, and thereby increase the commercial value of advertising. Our experiment result shows that our proposed approach performs better. Thoroughly, our method can not only enhance the average click rate of all ads in the website, but also make sure the fairness of exposure frequency of each ad. The experiment results demonstrate the effectiveness of our approach.